University of Wollongong

Research Online
Faculty of Science - Papers (Archive)

Faculty of Science, Medicine and Health

2007

Some approaches to new antibacterial agents

John B. Bremner
University of Wollongong, jbremner@uow.edu.au

Publication Details
Bremner, J. B. (2007). Some approaches to new antibacterial agents. Pure and Applied Chemistry, 79 (12), 2143-2153.

Research Online is the open access institutional repository for the

University of Wollongong. For further information contact the UOW
Library: research-pubs@uow.edu.au

Some approaches to new antibacterial agents

Abstract

Bacteria use a number of resistance mechanisms to counter the antibacterial challenge, and one of these is the
expression of transmembrane protein-based efflux pumps which can pump out antibacterials from within the
cells, thus lowering the antibacterial concentration to nonlethal levels. For example, in S. aureus, the NorA
pump can pump out the antibacterial alkaloid berberine and ciprofloxacin. One general strategy to reduce the
health threat of resistant bacteria is to block a major bacterial resistance mechanism at the same time as
interfering with another bacterial pathway or target site. New developments of this approach in the context of
dual-action prodrugs and dual-action (or hybrid) drugs in which one action is targeted at blocking the NorA
efflux pump and the second action at an alternative bacterial target site (or sites) for the antibacterial action
are discussed. The compounds are based on a combination of 2-aryl-5-nitro-1H-indole derivatives (as the
NorA efflux pump blocking component) and derivatives of berberine. General design principles, syntheses,
antibacterial testing, and preliminary work on modes of action studies are discussed
Keywords

Some, approaches, antibacterial, agents, CMMB

Disciplines

Life Sciences | Physical Sciences and Mathematics | Social and Behavioral Sciences
Publication Details

Bremner, J. B. (2007). Some approaches to new antibacterial agents. Pure and Applied Chemistry, 79 (12),
2143-2153.

This journal article is available at Research Online: http://ro.uow.edu.au/scipapers/3013

Pure Appl. Chem., Vol. 79, No. 12, pp. 21432153, 2007.
doi:10.1351/pac200779122143
2007 IUPAC

Some approaches to new antibacterial agents*

John B. Bremner
Department of Chemistry, University of Wollongong, Wollongong, NSW 2522,
Australia
Abstract: Bacteria use a number of resistance mechanisms to counter the antibacterial challenge, and one of these is the expression of transmembrane protein-based efflux pumps
which can pump out antibacterials from within the cells, thus lowering the antibacterial concentration to nonlethal levels. For example, in S. aureus, the NorA pump can pump out the
antibacterial alkaloid berberine and ciprofloxacin.
One general strategy to reduce the health threat of resistant bacteria is to block a major
bacterial resistance mechanism at the same time as interfering with another bacterial pathway
or target site. New developments of this approach in the context of dual-action prodrugs and
dual-action (or hybrid) drugs in which one action is targeted at blocking the NorA efflux
pump and the second action at an alternative bacterial target site (or sites) for the antibacterial action are discussed. The compounds are based on a combination of 2-aryl-5-nitro-1H-indole derivatives (as the NorA efflux pump blocking component) and derivatives of berberine.
General design principles, syntheses, antibacterial testing, and preliminary work on modes of
action studies are discussed
Keywords: antibacterials; efflux pumps; dual-action agents; prodrugs; berberine derivatives.
INTRODUCTION
Resistance to antibacterial agents by human pathogenic bacteria is an increasingly serious worldwide
health issue [13]. This resistance is apparent, for example, in the Gram-positive bacteria
Staphylococcus aureus and Enterococcus faecium, as well as in the Gram-negative pathogens
Escherichia coli and Pseudomonas aeruginosa, amongst others. The most pressing concerns are particularly with regard to the problematic human bacterial pathogens Acinetobacter baumanii, ESBL-producing Enterobacteriaceae, vancomycin-resistant E. faecium, P. aeruginosa, and methicillin-resistant
S. aureus (MRSA), as well as Aspergillus species of fungal pathogens [1]. The need for innovative and
efficient approaches to tackle this resistance problem, together with the judicious use of antibacterials,
is urgent.
Bacteria, no doubt due to exposure to natural antibiotics over millennia together with more recent
exposures to semi-synthetic and synthetic antibacterials, have evolved a number of mechanisms to
counter the antibacterial challenge. One major mechanism is based on enzymatic conversion of antibiotics into inactive forms, as typified by inactivation by the -lactamases of the penicillins and
cephalosporins through -lactam hydrolysis. A second mechanism, highlighted by resistance to vancomycin, is the modification of the biological target site. Replacement of the terminal D-alanine residue
in the cell wall peptidoglycan substrate for the cross-linking transpeptidase enzyme by D-serine or
D-lactate confers moderate and full resistance to vancomycin, respectively. This is mediated through reduction in binding of this antibiotic to the modified substrate [3].
*Paper based on a presentation at the 9th Eurasia Conference on Chemical Sciences, 913 September 2006, Antalya, Turkey.
Other presentations are published in this issue, pp. 21012177.

2143

J. B. BREMNER

2144

The third main mechanism involves reduction of intracellular antibacterial concentrations to sublethal levels. This may be mediated, for example, by reduced cell permeation or by active efflux mechanisms. Both Gram-positive and Gram-negative bacteria employ a range of such membrane pumps for
this purpose and invest a major effort in their production. For example, in S. aureus, the NorA pump is
a multidrug-resistance pump [4,5] which can extrude a range of antibacterial types, including the alkaloid berberine (2) and related quaternary salts, and fluoroquinolones, for example, ciprofloxacin (1).
The latter group also carries a positive charge through protonation at physiological pH. Thus, amphipathic compounds are characteristic substrates for this protein pump.

The effect of the pump can be demonstrated clearly using mutant strains of S. aureus. In a mutant
strain of S. aureus (K 2361), which over-expresses this NorA pump, the MIC (minimum inhibitory concentration) value for berberine is >650 M, while in the NorA-deleted mutant strain K1758, the corresponding value is much lower at 40 M and in wild-type S. aureus (8325-4) it is 325 M [6].
Gram-negative bacteria also employ a range of efflux pumps for protection against antibacterials,
including the clinically relevant RND pumps expressed in E. coli and P. aeruginosa [7,8].
Overall, the currently known bacterial efflux pumps have been categorized in four major families
as summarized in Table 1 [9]. These membrane-bound protein pumps vary in terms of size, numbers of
transmembrane segments, and general mode of action, as well as in terms of substrates. For the majority of them, proton motive force provides the energy source, while a minority in the ATP binding cassette family use ATP hydrolysis. Limited detailed structural information is available on these pumps although this is changing as more X-ray structural elucidations come to hand [10,11]. Similarly, the
detailed molecular mechanism involved in the operation of some of the pumps is starting to unfold
[12,13]. A particularly elegant example of this involving the use of detailed X-ray single crystal data
with the substrate antibiotic minocycline has been reported by Murakami and coworkers [12]. However,
the X-ray crystal structure of the NorA pump has not as yet been reported, although it is the focus of
much active research.
Table 1 Bacterial efflux pump classification.
Class
ABC superfamily
SMR family
MF family
MATE family
RND family

Energy
source

Substrate
specificity

Number of
amino acid residues

ATP hydrolysis
PMF
PMF
Na+
PMF

Specific, MDR
MDR
Specific, MDR
MDR
MDR

Variable
ca. 110
400600
ca. 450
Up to 1000

PMF: proton motive force; MDR: multidrug resistance; ABC: ATP binding cassette;
SMR: small multidrug resistance; MF: major facilitator; MATE: multi antimicrobial
extrusion; RND: resistance nodulation division.

2007 IUPAC, Pure and Applied Chemistry 79, 21432153

Approaches to new antibacterial agents

2145

In view of the importance of bacterial efflux pumps in mediating antibacterial resistance, they are
also significant targets for inhibition studies. Small molecule inhibitors of the pumps are capable of reversing antibiotic resistance, thus rendering antibiotics more effective again [9,14]. The multidrug-resistance NorA pump in S. aureus, the main focus of this paper, can be inhibited by a structurally diverse
range of small molecule inhibitors [9]. These inhibitor classes include indoles (e.g., INF55; [15,17]),
flavonolignans [16], and ureas (e.g., INF271 (4); [15,17]).

While the NorA inhibitor 5-nitro-2-phenyl-1H-indole (3) (INF55) has essentially no antibacterial
activity against wild-type S. aureus 8325-4 (MIC >525 M), if given in combination with berberine it
can decrease the MIC of the latter to 12.5 M [6].
DUAL-ACTION ANTIBACTERIAL STRATEGIES
These and other related observations suggested to us that further development of dual-action approaches
might result in a viable strategy for new antibacterial design, in which one action was specifically targeted at inhibition of an efflux pump. The second action could be directed to another bacterial target at
the same time.
In terms of generalized actions, three types of dual-action approaches can be delineated. The first
of these involve dual drugs, i.e., two drugs X' and Y' being administered with two different bacterial target sites A and B, respectively (Fig. 1). In the context of our studies, target site A is designated as an
efflux pump.

Fig. 1 Generalized dual-action strategies.

2007 IUPAC, Pure and Applied Chemistry 79, 21432153

2146

J. B. BREMNER

The second type of generalized approach is that of dual-action drugs or hybrid drugs (Fig. 1). In
this approach, one compound is involved, but with incorporated structural moieties of X and Y joined
by an enzymatically or chemically stable linker group (potentially, the moieties X and Y could also be
directly linked or structurally merged to afford chimeric molecules). Each moiety might then interact
separately with biological targets A and B (Fig. 1). It should be stressed, however, that the format pictured in Fig. 1 is the simplest one, and one could have, for example, cooperative binding at one or both
biological target sites A and B.
The third type of generalized approach is that of dual-action prodrugs. This is related conceptually to the hybrid design, but differs from it in that the linker would be designed to be cleaved enzymatically by bacterially specific enzymes to release both X' and Y', which could then interact with their
respective targets A and B (Fig. 1). While such compounds are related to tripartite prodrugs with a druglinker-carrier format, the carrier component is changed to be a second drug component, although it
could also act as a carrier component as well. Once again, it should be stressed that this represents the
simplest version of the dual-action prodrug approach, and more complicated variations on the theme
can be envisaged, for example, where X and Y and the cleavable linker are incorporated initially in a
cyclic skeleton.
Examples of all three general approaches in the antibacterial context have been reported in the literature (representative references include [1820] for dual drugs, dual-action prodrugs, and dual-action
drugs, respectively) but to our knowledge there were no specific examples prior to our reported work
[6] of antibacterial dual-action (hybrid) drugs, or of dual-action prodrugs, which incorporate an efflux
pump blocker as one component. Potential advantages of these approaches (over the administration of
two separate drugs) include synchronous, or near synchronous, delivery to different bacterial target
sites, as well as the possibility of a slower development of resistance. Disadvantages, however, include
high molecular weights and potential problems with synthesis, permeability issues, and differences in
the minimum efficacious concentrations required at the two different biological target sites. Another
disadvantage of dual-action drugs is the increased probability for reduced activity at one or both biological target sites for steric or electronic reasons, unless these molecules are carefully designed.
The focus of our research has been on dual-action antibacterials and dual-action prodrug antibacterials. Both types have been centered on a NorA efflux pump blocker component and an antibacterial component, where the antibacterial would normally be a substrate for the NorA pump (Fig. 2).

Fig. 2 General design principles for antibacterial dual-action drug/prodrugs incorporating NorA efflux pump
inhibition.

The initial realization of the design ideas was based on clues from nature. In a landmark study,
Lewis and Stermitz and coworkers reported [21] the co-occurrence of a NorA pump inhibitor
(5'-methoxyhydnocarpin) and an antibacterial substrate of the pump (berberine) in the North American
plant Berberis fremontii. The plant, which was observed to be free of bacterial disease, had thus developed a dual-action strategy to combat bacterial attack. While linkage of the hydnocarpin inhibitor and
berberine was considered, synthetic and stereochemical issues associated with the former compound
made this combination more difficult to achieve.
2007 IUPAC, Pure and Applied Chemistry 79, 21432153

Approaches to new antibacterial agents

2147

Simpler synthetic inhibitor compounds based on the 5-nitro-2-aryl-1H-indole skeleton had also
been reported to be NorA inhibitors [15], so our first-generation compounds were based on 5-nitro-2phenyl-1H-indole (3) (INF55), a good inhibitor of the NorA pump, together with berberine (2).

The dual-action prodrugs were based on linkage of the indole with berberine via ester (5a) or
amide (5b) linkages, while the dual-action drugs (6; R = H, 5'-OMe [Cl salt]) were based on a methylene group linkage with attachment to the 13 position of the berberine and the ortho-position of the
2-phenyl group.
Synthetically, both these chemical targets (5) and (6) were to be approached at the penultimate
stage using previously developed methodology [22] based on reaction of an appropriately substituted
indole-based alkyl bromide (the pump blocker component) with 8-allyldihydroberberine (7) (to provide
the berberine-based antibacterial component), the latter precursor component being accessed readily
from a nucleophilic addition of allyltributyl tin to berberine chloride (2) (Scheme 1).

Scheme 1

The synthesis of the indole precursors with the appropriate ortho-substituted phenyl rings was
based on a multistep synthesis, the first stage of which incorporated N-acylation of 5-nitroindole using
direct coupling with benzoic acid [23], palladium-mediated oxidative cyclization following the method
of Itahara [24], and hydrolytic ring-opening to give the acid (8) (Scheme 2) [17].

2007 IUPAC, Pure and Applied Chemistry 79, 21432153

2148

J. B. BREMNER

Scheme 2

Further functional group manipulations from the carboxylic acid (8) via the alcohol (9) provided
the benzyl bromide (10) and the bromo ester (11) (Scheme 3), as well as the amine (13) (via the azide
(12)), and the bromo amide (14) (Scheme 4), for the berberine coupling reaction as well as, for comparison purposes, the products to be expected from hydrolysis of the ester and amide prodrugs.

Scheme 3

Scheme 4

2007 IUPAC, Pure and Applied Chemistry 79, 21432153

Approaches to new antibacterial agents

2149

The final coupling reaction to give the ester (5a) (Scheme 5) proceeded in low yield. However, it
had the advantage of producing a final product at the oxidation level of berberine. The corresponding
amide analog (5b) of (5a) could also be produced by the same reaction. The mechanism of the process
probably follows that described previously [22] involving initial enamine C-alkylation in the dihydroberberine, followed by a [3,3]-sigmatropic rearrangement then a retro-ene reaction to give the isolated
product and elimination of propene; direct elimination of propene from the enamine alkylation product
cannot be excluded, however.

Scheme 5

The potential dual-action drug (15) was made analogously to the ester- and amide-linked derivatives and proceeded in better yield (Scheme 6) [6].

Scheme 6

2007 IUPAC, Pure and Applied Chemistry 79, 21432153

2150

J. B. BREMNER

INITIAL ANTIBACTERIAL TESTING RESULTS

Potential dual-action prodrugs
All the antibacterial testing was undertaken by Prof. K. Lewis, Mr. A. Ball, and Dr. G. Casadei from
Northeastern University, Boston, MA, USA, and a selection of results are given in this paper. Initially,
it was necessary to test the NorA pump inhibitory activity of the substituted indole alcohol (9) and the
corresponding amine (13), as they were the expected hydrolysis products from the respective ester- or
amide-linked prodrugs. The MIC values of (9) and of (13) in combination with a fixed sub-MIC concentration (80.7 M) of berberine in each case in S. aureus 8325-4 were 46.6 and 1.5 M, respectively;
the MIC value for (9) alone in S. aureus 8325-4 was 186.4 M, and for (13) it was 187.1 M. These
results were indicative of both (9) and (13) acting as pump inhibitors and thus potentiating the activity
of berberine in this bacterium, with the amine (13) showing considerable inhibitory potency [17].
The other expected prodrug hydrolysis product, the berberine carboxylic acid (16), which was
prepared separately from hydrolysis of the corresponding ethyl ester [25], was also assessed for direct
antibacterial activity (as its chloride salt) and was shown to have mild activity with an MIC of 116.3 M
(S. aureus 8325-4).

With the potential dual-action ester prodrug (5a), moderate activity was shown against the wildtype S. aureus strain 8325-4 (MIC 22.1 M), and, significantly, more potent activity was seen (MIC
4.3 M) against an S. aureus mutant strain (K1758) in which the NorA pump is deleted. While these
results are not inconsistent with (5a) acting as a dual-action prodrug, further evidence was needed that
hydrolysis was feasible. To this end, the ester (5a) was exposed to pig liver esterase but little hydrolysis to the alcohol (9) was seen (Scheme 7). When the ester (5a) at a very low, sub-MIC level was incubated with S aureus 8325-4 in Mueller Hinton broth and then a dichloromethane extract of the broth,
and of the bacterial pellet (after centrifugation of the culture broth), was analyzed by electrospray mass
spectrometry (ESMS) () and MS/MS, evidence for the presence of the alcohol (9) in low concentration was obtained. In determining the concentration of (9) by ESMS, the related methyl-substituted derivative (17) was used as an internal reference compound; the limit of detection of the alcohol (9) was
ca. 0.019 M. The alcohol hydrolysis product was chosen as the target for analysis as it was more readily extracted from aqueous media. In a separate control experiment, some hydrolysis of the ester by the
broth itself was also seen. Further work is thus necessary to assess the extent of bacterial hydrolysis of
(5a), looking at both the alcohol (9) and acid (16) hydrolysis products.
With the potential dual-action prodrug amide analog (5b) of (5a), the MIC value against S. aureus wild-type 8325-4 was 4.3 M.

2007 IUPAC, Pure and Applied Chemistry 79, 21432153

Approaches to new antibacterial agents

2151

Scheme 7

Potential dual-action antibacterial

High antibacterial potency was seen with the potential dual-action or hybrid agent (15; also designated
SS14). For example, against the Gram-positive bacteria S aureus 8325-4, E. faecalis V-583, and
Bacillus cereus T, MIC values of 3.1, 6.3, and 3.1 M were obtained, respectively. The corresponding
values for berberine itself were 325, >650, and 650 M. In further studies on (15; SS14), it was shown
[6] to rapidly accumulate in S. aureus cells, and also was considerably more potent than a mixture of
berberine itself and the NorA pump blocker INF55 (Fig. 3; reproduced with permission from [6]).
While (15) may not be a substrate for the NorA pump because of the bulky 13-indolyl substituent group,
it may, alternatively, be blocking its own efflux as initially designed. Further studies are planned to determine this in the future.
Antibacterial activity for (15) has also been demonstrated in vivo in the worm curing assay developed by Ausubel and coworkers [26] using the nematode Caenorhabditis elegans infected with
E. faecalis. In this assay, the berberine derivative (15) had a similar activity to tetracycline, and it was
as active as vancomycin. Interestingly, berberine itself showed poor in vivo activity in this assay as did
the pump blocker INF55, and the latter also showed toxic effects in the worms [6].
In initial studies on the potential mechanism of antibacterial action of (15), a comparison of DNA
binding of (15) and berberine (2) has been undertaken. These studies, which were based on ESMS (negative ion mode) and were undertaken by Ms. K. Gornall and Dr. J. Beck at the University of
Wollongong, confirmed that berberine (2) binds to double-stranded DNA [27], and also to triplex DNA,
but (15) does not bind to either of these two DNA forms [28]. However, (15) was shown to bind to
quadruplex DNA, as does berberine. The significance of this binding selectivity in the case of (15) is
being pursued currently.

2007 IUPAC, Pure and Applied Chemistry 79, 21432153

2152

J. B. BREMNER

Fig. 3 Potentiation of berberine action against S. aureus [6]. Actively growing S. aureus cells were treated with
berberine () or hybrid SS14 (). Berberine at a fixed concentration (1.87 M) was potentiated by varying amounts
of INF55 ().

CONCLUSION AND FUTURE PERSPECTIVES

Dual-action agents involving efflux pump inhibition hold considerable potential as agents to tackle certain drug-resistant human pathogenic bacteria. More work is needed in the area including the development of good ligand-based pharmacophores to inform more potent pump inhibitor design, as well as in
the development of dual-action agents and dual-action prodrugs targeting, in part, RND pumps in
Gram-negative bacteria. An X-ray structure on the NorA pump, the target of current work for example
by Prof. P. Henderson in the U.K., would also greatly advance the design and implemention of new
dual-action agents.
ACKNOWLEDGMENTS
I gratefully acknowledge the support and collaboration of coworkers at the University of Wollongong
(S. Samosorn, J. Ambrus, J. Copland, T. DuPree, L. Hick, J. Beck, K. Gornall, M. Kelso, J. Mbere, and
C. Perkins); Northeastern University, Boston (K. Lewis, A. Ball, G. Casadei); Harvard Medical School
(F. Ausubel, T. Moy); Southern IML Pathology, Wollongong (A. Clayton, R. Munroe); Avexa Avexa
Ltd., Melbourne (J. Deadman); and Srinakarinwirot University (S. Samosorn), Thailand. Support from
the Institute for Biomolecular Science, University of Wollongong is also acknowledged.
REFERENCES
1. G. H. Talbot, J. Bradley, J. E. Edwards, D. Gilbert, M. Scheld, J. G. Bartlett. Clin. Infect. Dis. 42,
657 (2006).
2. K. M. Overbye, J. F. Barrett. Drug Discov. Today 10, 45 (2005).
3. C. Walsh. Antibiotics: Actions, Origins, Resistance, ASM Press, Washington, DC (2003).
4. A. A. Neyfakh, C. M. Borsch, G. W. Kaatz. Antimicrob. Agents Chemother. 37, 128 (1993).
5. B. Marquez. Biochemie 87, 1137 (2005).
6. A. R. Ball, G. Casadei, S. Samosorn, J. B. Bremner, F. M. Ausubel, T. I. Moy, K. Lewis. ACS
Chem. Biol. 1, 594 (2006).
7. H. Nikaido. In Frontiers in Antimicrobial Resistance: A Tribute to Stuart. B. Levy, D. G. White,
M. N. Alekshun, P. F. McDermott (Eds.), pp. 261274, ASM Press, Washington, DC (2005).
8. O. Lomovskaya, H. Zgurskaya, M. Totrov, W. J. Watkins. Nat. Rev. Drug Discov. 6, 56 (2007).
2007 IUPAC, Pure and Applied Chemistry 79, 21432153

Approaches to new antibacterial agents

2153

9. G. W. Kaatz. Curr. Opin. Invest. Drugs 6, 191 (2005).

10. S. Murakami, R. Nakashima, E. Yamashita, A. Yamaguchi. Nature 419, 587 (2002).
11. M. K. Higgins, E. Bokma, E. Koroakis, C. Hughes, V. Koronakis. Proc. Natl. Acad. Sci. USA 101,
9994 (2004).
12. S. Murakami, R. Nakashima, E. Yamashita, T. Matsumoto, A. Yamaguchi. Nature 443, 173
(2006).
13. M. A. Seeger, A. Schiefner, T. Eicher, F. Verrey, K. Diederichs, K. M. Pos. Science 313, 1295
(2006).
14. J.-M. Pages, M. Masi, J. Barbe. Trends Mol. Med. 11, 382 (2005).
15. P. N. Markham, E. Westhaus, K. Klyachko, M. E. Johnson, A. A. Neyfakh. Antimicrob. Agents
Chemother. 43, 2404 (1999).
16. N. R. Guz, F. R. Stermitz. J. Nat. Prod. 63, 1140 (2000).
17. S. Samosorn, J. B. Bremner, A. Ball, K. Lewis. Bioorg. Med. Chem. 14, 857 (2006).
18. L. E. Kehoe, J. Snidwongse, P. Courvalin, J. B. Rafferty, I. A. Murray. J. Biol. Chem. 278, 29963
(2003).
19. J. M. T. Hamilton-Miller. J. Antimicrob. Chemother. 33, 197 (1994).
20. C. Hubschwerlen, J.-L. Specklin, D. K. Baeschlin, Y. Borer, S. Haefeli, C. Sigwalt, S. Schroeder,
H. H. Locher. Bioorg. Med. Chem. Lett. 13, 4229 (2003).
21. F. R. Stermitz, P. Lorenz, J. Tawara, L. A. Zenewicz, K. Lewis. Proc. Natl. Acad. Sci. USA 97,
1433 (2000).
22. J. B. Bremner, S. Samosorn. Aust. J. Chem. 56, 871 (2003).
23. J. B. Bremner, S. Samosorn, J. I. Ambrus. Synthesis 2653 (2004).
24. T. Itahara. Heterocycles 24, 2557 (1986).
25. S. Samosorn. Ph.D. thesis, University of Wollongong (2005).
26. T. I. Moy, A. R. Ball, Z. Anklesaria, G. Casadei, K. Lewis, F. M. Ausubel. Proc. Natl. Acad. Sci.
USA 103, 10414 (2006).
27. S. Mazzini, M. C. Belluci, R. Mondelli. Bioorg. Med. Chem. 11, 505 (2003).
28. K. C. Gornall, S. Samosorn, J. Talib, J. B. Bremner, J. L. Beck. Rapid Commun. Mass Spectrom.
21, 1759 (2007).

2007 IUPAC, Pure and Applied Chemistry 79, 21432153